Maria coordinates a doctoral school of sixty researchers, and she spends her professional life applying a standard she learned in her own research years: no methods, no belief. When a manuscript lands on a reviewer’s desk claiming a result without describing how it was measured, the review is short. Yet every month, training proposals land on Maria’s desk making claims about stress, resilience and performance, and where the methods section should be, there is a carousel of testimonials. Maria is asked to fund, from a budget she defends line by line, the kind of claim she would reject in thirty seconds as a referee. This article is the manuscript I believe she is owed: what Paul and I measure in the Resilience Training, what changed in a real cohort, what did not, and where the limits of the claim sit. Peer review welcome.
Table Of Contents:
- The methods section
- Where the cohort started: above the field’s crisis line
- The results section
- What did not move, and why the nulls are load-bearing
- Limitations
- What Maria asks now
- FAQ
- References
The methods section
Every Resilience Training cohort completes a full battery in week one and again in week eight: the Perceived Stress Scale, the Hospital Anxiety and Depression Scale, the Positive and Negative Affect Schedule, the Satisfaction with Life Scale, a job satisfaction measure, working hours, the WHO performance items, and the Personal Resilience Indicator, the instrument Paul and I created and I validated on eight hundred and three working professionals. Not a feedback form. The same class of instruments the doctoral wellbeing literature itself runs on, which is deliberate: it makes our numbers comparable to the field’s.
The cohort below is February 2026: thirty five researchers with matched before and after data, doctoral students, postdocs and group leaders from institutions including Max Planck institutes and CERN. Ten researchers completed only the first questionnaire, so we checked whether the dropouts differed at baseline from the completers. They did not, on any measure. That check matters more than it looks: without it, every improvement below could be an artefact of the people who felt better being the people who returned. If you want the background before the results, my article on what resilience training involves carries it.
Where the cohort started: above the field’s crisis line
The doctoral wellbeing literature has settled on a grim consensus number. Roughly one in three early-career researchers carries elevated psychological distress: the Levecque study put PhD students at one in three risk of a common psychiatric disorder [1], the Evans survey found anxiety and depression at several times the general population rate [2], and the 2026 meta-analysis of one hundred and thirty eight thousand early-career researchers landed at thirty per cent [3].
Our cohort arrived above that line. At week one, sixty nine per cent of the researchers scored in the at-risk range for anxiety on the screening scale, and thirty four per cent in the at-risk range for low mood. Screening figures, not diagnoses; load, not illness. But read the first number again, because it is the state in which a training cohort actually walks through the door: two researchers in three carrying an anxiety load the instrument classifies as worth clinical attention. Whatever a training claims to do, this is the population it has to do it for. If that population sounds like you rather than like a statistic, my article on whether you need resilience training is the place to take the question.
The results section
Every one of the seven wellbeing scales improved, with medium to large effects, and every improvement survived two independent statistical tests. Perceived stress fell from 20.7 to 15.1, from far above the general adult norm of around thirteen to fifteen to just beside it. Anxiety fell from 10.1 to 6.4, carrying the average researcher from above the at-risk threshold to below it. Low mood fell from 6.5 to 3.6. Positive affect, life satisfaction and job satisfaction rose; negative affect fell.
The threshold movements are the findings I would put in the abstract. The at-risk share for anxiety halved, sixty nine per cent to thirty four per cent: twelve researchers moved out of the at-risk range and not one moved into it. For low mood, thirty four per cent to six per cent, ten out, none in. In eight weeks, a cohort that arrived above the field’s crisis line finished at it on anxiety and well under it on low mood.
Resilience itself, on the PRI, rose from 3.29 to 3.85 on the five point scale, with all six domains moving: Health, Purpose, Cognition, Perseverance, Composure and Relationships. Health rose most, and the sleep data explains why: sleep debt across the cohort roughly halved, from 5.8 hours a week to 2.9, and sleep efficiency climbed from eighty two to ninety one per cent, across the threshold sleep researchers treat as the marker of a well-functioning night. The mechanics of that connection live in my post on sleep and resilience.
What did not move, and why the nulls are load-bearing
Working hours did not change. Forty one hours a week before, forty after, statistically indistinguishable, and the WHO performance items complete the picture from the other side: researchers’ rating of their own performance on the days they worked rose from 6.2 to 7.4 on the ten point scale, a medium effect, on those unchanged hours. Self-rated, so a perception measure, and I say so. But hold the pair together, because it dissolves the objection every coordinator carries into this conversation. The cohort did not buy its wellbeing with research time. It reported feeling better and working better on the same working week.
Two more results stayed still, and they stay in the paper. Of the twelve resilience drivers, Emotional Literacy did not move meaningfully and Intuition moved only a little. Eight weeks does not rebuild everything, and a results table where everything improves is indistinguishable from a results table that measures nothing. The ten drivers that moved are credible because the two that did not are printed beside them.
Limitations
A training cohort is not a controlled trial. There is no control group, so I do not write that the training caused these changes; I write that researchers who completed it showed them. February to spring carries a seasonal tailwind for mood, which is why the least seasonal measures, job satisfaction, working hours and the resilience scores, carry the most weight here. The anxiety and low mood figures are screening results and are never called anything else. These limits are real, they are stated, and they are also the same limits every provider in this field faces. The difference is that, to my knowledge, we are the only one publishing the measurements at all. The field counts the problem in meticulous detail. Almost nobody measures whether anything helps.
What Maria asks now
Maria’s procurement standard, translated from peer review into three questions any provider should survive: What do you measure? When do you measure it? And will you show me what did not move? A provider with answers is selling a return. A provider without them is selling testimonials, and Maria has a referee’s report for that. The full cohort data behind this article is available to coordinators on request, methods and limitations included, and Paul and I will walk through any of it in as much depth as you want. A participant’s week by week account of the training itself follows in a coming piece, and how the resilience measurement works in practice is here.
FAQ
Does resilience training actually work?
Here is what nobody tells you: most providers cannot answer this question, because they never measure. In our February 2026 cohort, measured with validated instruments in week one and week eight, every wellbeing scale improved with medium to large effects, the share of researchers in the at-risk range for anxiety halved, self-rated performance rose while hours stayed flat, and resilience rose across all six domains of the Personal Resilience Indicator. No control group, so we describe change rather than claim cause. Ask any provider what they measure and when. The silence is your answer.
How is the effect of resilience training measured?
With the same instruments the research literature uses: the Perceived Stress Scale, the Hospital Anxiety and Depression Scale for screening anxiety and low mood, the Positive and Negative Affect Schedule, the Satisfaction with Life Scale, a job satisfaction measure, working hours, WHO performance items, sleep metrics, and the Personal Resilience Indicator, completed by the same person before the training and at its end. Matched pairs are what make a comparison mean something. Everything else is a satisfaction score wearing a lab coat.
Will the training reduce research output?
The measured answer is no. In our cohort, working hours did not change between week one and week eight, while stress, anxiety and low mood all fell and researchers’ self-rated performance on the days they worked rose from 6.2 to 7.4 on the ten point WHO scale. The gains did not come from working less. They came from better recovery, better sleep and better structure around the same working week, and the cohort rated itself as working better for it.
References
[1] Levecque, K., Anseel, F., De Beuckelaer, A., Van der Heyden, J., Gisle, L. (2017). Work organization and mental health problems in PhD students. Research Policy, 46(4), 868 to 879.
[2] Evans, T. M., Bira, L., Gastelum, J. B., Weiss, L. T., Vanderford, N. L. (2018). Evidence for a mental health crisis in graduate education. Nature Biotechnology, 36, 282 to 284.
[3] Dreisoerner, A., Goetz, V., Frohnmayer, D., Tran, U. S., Voracek, M., Nater, U. M. (2026). Prevalence and severity of mental health problems in early-career researchers: a systematic review and meta-analysis. Nature Human Behaviour.
Last Updated on August 12, 2026 by Dr Nadine Sinclair
Dr Nadine Sinclair
Dr. Nadine Sinclair is a molecular biologist, strategy consultant and co-founder of Mind Matters. A scientist by training and at heart, she conducted her doctoral research at the Max Planck Institute for Biophysical Chemistry. She has 30,000+ hours of hands-on project management experience, built across 18 years as a strategy consultant at McKinsey and as an independent consultant, working with research institutions, foundations, pharmaceutical and biotech companies and governments. She is the co-creator of the Personal Resilience Indicator and the author of On Track. Since 2018, over 1,500 researchers from more than 30 institutions have trained with Mind Matters.